As a key technology for assisting driving, the detection of moving targets can warn vehicle drivers of the possibility of collisions with nearby obstacles and provide powerful technical support for avoiding traffic accidents. However, when traditional motion detection algorithms are used in the detection of moving objects in automobiles, it is difficult to ensure real-time performance in complex scenes and high detection accuracy. To solve this problem, this paper proposes a moving object detection algorithm based on RPCA. Firstly, using the maximum similarity between two adjacent frames in the surveillance video, this similarity is expressed as an observation matrix with low rank, and then the low-rank background in the video is recovered from the matrix. Secondly, because there is a lot of redundant information between adjacent frames, the improvement of the construction of RPCA observation matrix reduces the spatial complexity of the observation matrix. Finally, the frame difference method is combined with RPCA to detect the moving target, and the significant area of the moving target is improved. The experimental results show that the algorithm achieves good experimental effect in different scenarios and can accurately extract the moving objects in the video. Compared with the classic three-frame difference method, background subtraction method and traditional method based on RPCA, the detection effects were improved by 54.32%, 36.63% and 2.73%, respectively. Moreover, by reducing the space complexity of the RPCA algorithm, the complexity of the detection algorithm is reduced, thereby reducing the energy consumption of the automobile, reducing the carbon emission, and improving the environmental protection and economy of the automobile.